This is an automated email from the ASF dual-hosted git repository.
ruifengz pushed a commit to branch branch-4.0
in repository https://gitbox.apache.org/repos/asf/spark.git
The following commit(s) were added to refs/heads/branch-4.0 by this push:
new fab541d43395 [SPARK-51140][ML] Sort the params before saving
fab541d43395 is described below
commit fab541d43395a61c1b295aa46717d183bf4236ff
Author: Ruifeng Zheng <[email protected]>
AuthorDate: Tue Feb 11 08:10:20 2025 +0800
[SPARK-51140][ML] Sort the params before saving
### What changes were proposed in this pull request?
Sort the params before saving
### Why are the changes needed?
to improve debugability:
when developing ml connect, sometime I need to manually check the stored
models, I notice that
the params are always saved unsorted, make it hard to compare the params:
before:
```
{"class":"org.apache.spark.ml.clustering.PowerIterationClustering","timestamp":1738926090947,"sparkVersion":"4.1.0-SNAPSHOT","uid":"PowerIterationClustering_a5c66eecaec6","paramMap":{"dstCol":"dst","k":2,"weightCol":"weight","maxIter":40,"srcCol":"src","initMode":"random"},"defaultParamMap":{"dstCol":"dst","k":2,"maxIter":20,"srcCol":"src","initMode":"random"}}
```
```
{"class":"org.apache.spark.ml.clustering.PowerIterationClustering","timestamp":1738926386839,"sparkVersion":"4.1.0-SNAPSHOT","uid":"PowerIterationClustering_b91c5734d913","paramMap":{"k":2,"initMode":"random","weightCol":"weight","srcCol":"src","maxIter":40,"dstCol":"dst"},"defaultParamMap":{"k":2,"initMode":"random","srcCol":"src","maxIter":20,"dstCol":"dst"}}
```
after:
```
{"class":"org.apache.spark.ml.clustering.PowerIterationClustering","timestamp":1739154410677,"sparkVersion":"4.1.0-SNAPSHOT","uid":"PowerIterationClustering_483e02530367","paramMap":{"k":2,"maxIter":40,"weightCol":"weight"},"defaultParamMap":{"dstCol":"dst","initMode":"random","k":2,"maxIter":20,"srcCol":"src"}}
```
### Does this PR introduce _any_ user-facing change?
no
### How was this patch tested?
existing tests and manually check
### Was this patch authored or co-authored using generative AI tooling?
no
Closes #49861 from zhengruifeng/ml_store_params.
Authored-by: Ruifeng Zheng <[email protected]>
Signed-off-by: Ruifeng Zheng <[email protected]>
(cherry picked from commit 8071868d1fc9c78a46d0095e38b450e88abcd4e6)
Signed-off-by: Ruifeng Zheng <[email protected]>
---
mllib/src/main/scala/org/apache/spark/ml/util/ReadWrite.scala | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/mllib/src/main/scala/org/apache/spark/ml/util/ReadWrite.scala
b/mllib/src/main/scala/org/apache/spark/ml/util/ReadWrite.scala
index dcb337218edc..d155f257d230 100644
--- a/mllib/src/main/scala/org/apache/spark/ml/util/ReadWrite.scala
+++ b/mllib/src/main/scala/org/apache/spark/ml/util/ReadWrite.scala
@@ -486,8 +486,8 @@ private[ml] object DefaultParamsWriter {
paramMap: Option[JValue]): String = {
val uid = instance.uid
val cls = instance.getClass.getName
- val params = instance.paramMap.toSeq
- val defaultParams = instance.defaultParamMap.toSeq
+ val params = instance.paramMap.toSeq.sortBy(_.param.name)
+ val defaultParams = instance.defaultParamMap.toSeq.sortBy(_.param.name)
val jsonParams = paramMap.getOrElse(render(params.map { case ParamPair(p,
v) =>
p.name -> parse(p.jsonEncode(v))
}.toList))
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
